Valency rearrangement phenomena and their limitations with Polish ‘clear’
verbs1
Anna Malicka-Kleparska, dr hab., prof. KUL JP II
John Paul II University of Lublin
Department of English
Al. Racławickie 14,
20-950 Lublin
adres domowy:
ul. Smyczkowa 4/139
20-844 Lublin
tel. 606983641
e-mail: malic@kul.lublin.pl
Summary: The paper is concerned with very basic questions concerning the nature of lexical
representation of verbs which take part in valency rearrangement alternations and the
contribution of grammar to the morpho-syntactic phenomena involved in this
rearrangement.
We analyze a group of such valency rearrangement verbs which are referred to as ‘clear’
verbs in Polish and compare their behavior and properties with the corresponding material
from English, Greek and Hebrew. In particular we try to establish which properties enable
some verbs to undergo the alternation, while some others accept only one of the alternants.
We propose lexical representations for alternating and non-alternating verbs and establish a
connection between these representations and the alternating propensity of prefixed verbs
in Polish.
Streszczenie artykułu: Przeszeregowania walencji i ograniczenia na nie nałożone w grupie polskich
czasowników dotyczących ‘czyszczenia’
Artykuł daje odpowiedzi na bardzo podstawowe pytania dotyczące reprezentacji leksykalnej
czasownika i podziału zadań pomiędzy tę reprezentację a operacje gramatyczne w odniesieniu do
morfo-syntaksy czasowników, których argumenty wewnętrzne mogą być realizowane różnorodnie.
Analizuję tu grupę polskich czasowników dotyczących czyszczenia, które przydzielają dwa argumenty
wewnętrzne, w formie dopełnienia wyrażonego frazą rzeczownikową w bierniku i przyimkową frazą
z elementem głównym z, w przypadku których zamianie struktury syntaktycznej nie towarzyszy
zmiana znaczenia. Zastanawiam się, co jest przyczyną takiego stanu rzeczy, że tylko część
czasowników o odpowiednim znaczeniu pozwala na tę alternację. Jednocześnie staram się ustalić
1
The first version of this paper, entitled Deprivation verbs in Polish, was presented at SinFonIJa 5 in Vienna, 27th
September 2012. The discussion that issued there and then made me re-think and revise some major claims of
the original, for which I would like to express my thanks to the Participants and especially to prof. Rok Žaucer
from the University of Nova Gorica. An improved version of the paper was reviewed insightfully by prof. Anna
Bondaruk and prof. Bożena Cetnarowska, who also offered valuable comments. Of course all the remaining
shortcomings of this paper are entirely my responsibility.
jaka reprezentacja leksykalna może wyjaśnić zróżnicowane zachowanie tych czasowników i jakie
towarzyszą jej zjawiska gramatyczne. Znajduję związek pomiędzy elementami prefiksalnymi w takich
czasownikach, a ich właściwościami alternacyjnymi. Dane dla języka polskiego porównuję z danymi z
angielskiego, greckiego i hebrajskiego.
Key words: morpho-syntax, verbal valency, lexicon, event structure, Polish, English, Geek, Hebrew
Słowa-klucze: morfo-syntaksa, walencja czasownikowa, leksykon, struktura zdarzenia, polski,
angielski, grecki, hebrajski
1. Introduction of the data
Recently a lot of attention in morpho-syntax has been devoted to such verbs which show valency
rearrangement possibilities of the type where the two internal arguments of a single verb adopt
alternative positions without any major change of meaning.2 The subclass of such verbs analyzed in
this paper is referred to in linguistic literature as ‘clear’ verbs, as they, generally speaking, convey the
meaning that something is removed, i.e. ‘cleared’, from a Location, or, in the other variant ,some place
is ‘cleared of ‘ some Substance. In Polish these verbs show little researched irregularities in their
morpho-syntax: many allow this re-arrangement, as the examples in (1) below illustrate; others can
function in a specific frame only: (2) below illustrates such situations where the verb is followed by a
Noun Phrase in the Accusative case (Direct Object) specifying Substance (Stuff) and by a
Prepositional Phrase introduced by the preposition z- plus an NP in the Genitive case – informing
about Location. Examples in (3) contain the third subclass of ‘clear’ verbs which admit of the opposite
frame – DO Location and Stuff in the PP. The type of morpho-syntactic structure in (2) is called
Change of Location Frame (COL), while the one in (3) – Change of State (COS).
Many Polish ‘clear’ verbs associate with both COL and COS structures. Such predicates are
called alternating ‘clear’ verbs. A selected sample is supplied in (1) below with the relevant
internal arguments typed in bold characters:3
1.
a. czyścić ‘clean’
COL - Narzędziem skutecznie czyszczącym dane z dysków jest program Acronis DriveCleanser.’ The
instrument effectively cleaning data off disks is Acronis DriveCleanser’
COS - Czyścili podłogę z krwi. ’They cleaned the floor of blood’
b. szorować ‘scrub’
2
See e.g. Rappaport Hovav and Levin (1998), Levin (2006) for English, Segal and Landau (2012) for Hebrew,
Alexiadou and Anagnostopoulou (2012) for Greek.
3
The data, whenever possible, have been taken from the National Corpus of the Polish Language, included in
the references as: Przepiórkowski, Bańko, Górski and Lewandowska-Tomaszczyk (2012). This has been done to
avoid notorious difficulties with grammaticality judgments that appear with this material. In further sections of
this paper the author’s own examples will also be given as necessitated by the complexity of the structures
since we will have to supply the appropriate testing ground: Linguistic corpora frequently fail to offer such
intricate phrases, especially when we aim at particular lexical elements, not at numerous classes. Whenever a
reasonable doubt may arise whether the data are taken from the Corpus or supplied by the author of this
paper, such data are initialed: (A.M.).
COL - Kobiety szorowały brud z podłogi. (A.M.) ‘Women scrubbed dirt off the floor’
COS - Kobiety wiejskie szorowały naczynia z resztek mięsa i tłuszczu. ‘Country women scrubbed dishes
from the remaining meat and fat’
c. sprzątać ‘clean’
COL - Sprzątali śmieci z ulic. ’They cleaned trash off the streets’
COS - Co rano trzeba było sprzątać okolicę z prezerwatyw. ‘Every morning it was necessary to clean
the surroundings of condoms’
d. drenować ‘drain’
COL - drenując pieniądze z kieszeni podatników ‘draining money from the pockets of tax-payers’4
COS - poza handlowcami, drenującymi rynek z pieniędzy ‘apart from businessmen, draining the
market of money’
e. grabić ‘rob’
COL - Grabiąc złom z posesji, dokonują dużej szkody. ‘By robbing rummage from the property they do
a lot of damage’
COS – Grabiąc posesję ze złomu, dokonują dużej szkody. (A.M.) ‘By robbing the property of rummage
they do a lot of damage’
As the examples above show, both frames convey the same meaning and both are equally
grammatical. Polish has a number of alternating verbs, but for the purposes of our exposition
we have selected mostly such that occur in the Corpus in the relevant frames.
Then some other verbs, with the same semantics, appear in the COL frame exclusively, while
COS with them is not merely unattested, but ungrammatical, e.g.:
2.
a.
łamać ‘break’
COL - Łamię gałązkę z rosnącego przy drodze bzu. ‘I am breaking a twig from a lilac bush
growing near the road’
COS – *Łamię rosnący przy drodze bez z gałązki. (A.M.) *’I am breaking a lilac bush growing near the
road of a twig’
b.
ratować ‘rescue’
COL - Uległ tragicznemu wypadkowi ratując dzieci z pożaru. ‘He had a tragic accident while rescuing
the children from the fire’
c.
usunąć ‘remove’
COL - Będą usuwać bezdomnych z dworców. ‘They will remove the homeless from railway-stations’
e.
brać ‘take’
COL - Brali pieniądze z miejskiej kasy. ‘They took money from the city funds’
f.
zdejmować ‘take off’
COL - Zdejmują nogę z gazu. ‘They take off their feet from the accelerators’
Finally, there are ‘clear’ verbs that favor the COS frame, while COL is inadmissible with them,
e.g.:
3.
a. leczyć ‘cure’
COS - Ja nie ośmieliłbym się leczyć kogoś z depresji. ‘I would not dare to treat anybody for depression’
4
English translations of the data are frequently very awkward, but we strive to stick as much as possible to the
Polish original pattern, which – as it will become very clear in the course of this paper – diverges from what
English allows. In a nutshell it has few alternating verbs, while Polish has dozens of them.
COL – *Ja nie ośmieliłbym się leczyć depresję z kogoś. *’I would not dare to treat depression from
anybodyody’
b. plądrować ‘plunder’
COS - Gdy jedna z kobiet zajmowała się osobą pokrzywdzoną, druga plądrowała mieszkanie z
pieniędzy. ‘While one of the women was taking care of the hurt person, the other plundered the
appartment from money’
c. otrzeźwić ‘sober’
COS – Otrzeźwili ojca z upojenia. ‘They have sobered the father from intoxication’ (A.M.)
d. opróżniać ‘empty’
COS - Nie można jednocześnie opróżniać szafy ze5 starych, nierzadko zniszczonych ubrań. ‘You cannot
empty the closet of old, frequently damaged, clothes’
e. obrobić ‘rob’
COS - Obrobili mieszkanie z cennych przedmiotów i pieniędzy. ‘They robbed the appartment of
precious objects and money’
Some ‘clear’ verbs appear with one full frame and the other truncated, i.e. shortened by the
phrasal argument6 :
4.
a.
COL–only verb - Addy wchodzi do pokoju pełnego butelek Smirnoffa, odkręca
zakrętkę z jednej z nich. ‘Addy enters the room full of bottles of Smirnoff vodka, unscrews the cap
from one of them’
Truncated COL - Addy odkręca butelkę..‘Addy unscrews the bottle’
Truncated COS - Addy odkręca zakrętkę. ‘Addy unscrews the cap’
COS - *Addy odkręca butelkę z zakrętki. *‘Addy unscrews the bottle of the cap’
b. CO –only verb - leczyć kogoś z depresji ‘treat sb. for depression’
Truncated COL – leczyć kogoś ‘treat sb.’
Truncated COS – leczyc depresję ‘treat depression’
COL - *leczyć depresję z kogoś *‘treat depression from sb.’
The alternating verbs allow also truncated frames in all the cases, e.g.: czyścić dane, czyścić dysk (see
1.a.), however COL-only verbs (see 2. above) and COS –only verbs (see 3. above)are unpredictable in
this respect as the examples in (5.) below, when compared with these in (4.), show:
5.
a. COL-only verb: ratować dzieci z pożaru (see 2.b)
Truncated COS - ratować dzieci
Truncated COL - *ratowac pożar
b. COS-only verb: plądrowała mieszkanie z pieniędzy (see 3.b)
Truncated COL – plądrowały mieszkanie
Truncated COS - *plądrowały pieniądze
Summing up, in Polish certain ‘clear’ verbs allow the valency re-arrangement alternation and thus
can appear either in COS and COL frames indiscriminately, as well as in truncated COS and COL
frames. Contrariwise, COL-only and COS-only verbs appear in truncated COS and COL frames
respectively, but their appearance in the truncated frame of the opposite type cannot be predicted,
5
The preposition z has an allomorphic variant ze-. For details of this allomorphy see e.g. Gussmann (2007:235).
6
The term ‘truncated frame’ is taken from Segal and Landau (2012).
just like whether a given ‘clear’ verb will alternate or not. This situation resembles the data in English,
Greek or Hebrew, to be discussed in section 3. In Polish, however, there may exist a certain bond
between verbal morphology and morpho-syntax, so the ‘clear’ alternation phenomena may not be
quite as random as in other languages. This issue will be substantiated on the basis of prefixed verbs
in the next section.
2. Valency rearrangement phenomena with prefixed ‘clear’ verbs in Polish
Such a complicated and highly idiosyncratic body of data breeds many questions concerning the
nature of their representation in the Lexicon, the type of information which should be available
therein as well as to the contribution made by grammatical, regular structures/principles or
processes.
So far we have presented the data as random in the choice of the frame variant. Nevertheless, Polish
shows a certain propensity towards the alternation if verbs are prefixed. Below we will illustrate the
situation with examples of prefixed verbs that show the ‘clear’ alternation, while the non-prefixed
variants do not have the same force. A few prefixes have this effect. Sometimes both frames are
allowed, but occasionally only one or the other. Below we will give examples with wy- (in 6 below), z/s-/ś-7 (7), ob- (8), od- (9), o- (10), where the relevant data are the most prominent, and also some
incidental cases in (11).
6.
a. wycisnąć ‘squeeze’8
COL - […] maszynę, która wyciśnie olej z nasion. ‘a machine which will squeeze oil from seeds’
COS - […] wycisnę tubkę z połowy [zawartosci]! ‘I will squeeze the tube out of half [of its content]’
cisnąć9 ‘throw’, e.g. Ktoś inny cisnął miskę z rozmachem przed siebie.‘Sb. else threw the bowl ahead
with force’
b. wytrzepać ‘beat, shake ‘
COL – wytrzepał słomę z butów ‘He shook out straw from his shoes’
COS - Tu trzeba ratować ziemską cywilizację, wytrzepać Planetę z pierza. ‘Here one must save the
Earth civilization, beat the Planet from fluff’
trzepać ‘beat’
COL – trzepać kurz z dywanu (A.M.) ‘to beat a carpet from dust’
c.wycierać ‘wipe’
COL - wyciera kredę z dłoni ‘He wipes chalk off his hand’
COS - wyciera wąsy z musztardy ‘he wipes his moustache of mustard’
trzeć ‘rub’, e.g. trzeć ręce ‘rub hands’
d. wysuszyć ‘dry’
COL - Wysusz wodę z włosów! ‘Dry water from your hair’
COS - Wysusz włosy z wody! ‘Dry your hair of water’
suszyć ‘dry’, e.g. COS - Susz włosy z wody! ‘Dry your hair of water’
e. wyprać ‘wash up’
COL - wyprać plamy z sukni ‘wash stains off a dress’
COS - wyprać sumienia z poglądów politycznych ‘wash conscience.pl from political views’
prać ‘wash up’
COS – prać bieliznę z brudu (A.M.)’wash the underware from dirt’
f. wyprzedawać ‘sell out’
7
See Gussmann (2007:210) for the alternations involved.
Prefixes are indicated with bold characters.
9
Whether the verbs cisnąć and wycisnąć are really morphologically related may be disputed. Prefixation in
Polish with verbs introduces a variety of meanings into the verbal stems, frequently only partially predictable,
or utterly idiosyncratic. We believe the verbs we have chosen for our analysis show close enough similarity of
meaning and form to be believed related. For recent accounts on verbal prefixation in Polish see Szymanek
(2010), Wiland (2011), for a general panorama of Slavic prefixation see Svenonius (2004).
8
COL – wyprzedawała mienie z domu (A.M.) ‘She sold out the belongings from home’
COS - wyprzedawała rodzinę z mienia ‘She sold her family out of their belongings’
sprzedawać ‘sell’10
COL – sprzedawała mienie z domu (A.M.) ‘She sold the belongings from home’
g. wywietrzyć ‘air’
COL - wywietrzyć gaz z mieszkania ‘to remove gas from the flat’
COS - wywietrzyć mieszkanie z nagromadzonego gazu ‘air the flat free from gas’
wietrzyć ‘air’
COS – wietrzył mieszkanie z gazu ‘He aired the flat free from gas’
h. wyciągać ‘pull out’
COL - wyciągać piłkę z siatki ‘to pull out a ball from the net’
ciągnąć ‘pull’, e.g.: ciągnąć sieci ‘pull the nets’
7.
a. spłukać ‘flush, rinse’
COL - Deszcz spłukał farbę z tabliczek .‘Rain has rinsed the paint from tablets’
COS - spłukać ręce z tuszu i kleju ‘to rinse hands from ink and glue’
płukać.
b. zrzucać ‘throw down’, wyrzucać ‘throw away’
COL - Przy większych opadach należałoby zrzucać śnieg z dachów. ‘After a bigger snowfall one should
throw down snow from roofs’
COL - Wyrzucali wszystko z meblościanki. ‘They threw everything away from a set of shelves’
rzucać ‘throw’, e.g. Rzucał plecak. ‘He threw his rucksack’
c. zmyć ‘wash’
COL - zmywający farbę z policzków wojowników ‘washing out pain from the warriors’ cheeks’
COS - zmywając ciało z grzechu ‘washing the body of sin’
myć ‘wash’
COS - on mył wannę z czarnych plam ‘He washed a bath-tub of black stains’
d. ścierać ‘wipe out’
COL - ścierać wodę ze ścian ‘to wipe water from the walls’
COS - ścierała podłogę z krwi ‘She wiped the floor of blood’
trzeć ‘rub’, e.g. Tarł rękę ‘He rubbed his hand’
e. zsuwać ‘push’
COL - zsuwasz kamień z pokrywy ‘You push a stone from the lid’
suwać ‘push’, e.g.: suwać nogami ‘to drag one’s feet’
f. zdmuchnąć ‘blow out’
COL - zdmuchnęło czapkę z głowy ‘It blew out a hat off one’s head’
dmuchnąć ‘blow’, e.g. dmuchnąć w trąbkę ‘to blow a trumpet’
g. zedrzeć ‘scratch, tear’
COL - zdarł dach z budynku ‘It tore the roof off a building’
drzeć ‘tear’, e.g. drzeć papier ‘to tear paper’
h. zgarniać 11‘gather’
COL - zgarniała piasek z chodnika
i. zdrapać ‘scratch’
COL - zdrapać folię z kart ‘to scratch foil from cards’
drapać ‘scratch’, e.g.: drapać rękę ‘to scratch one’s hand’
j. zbierać ‘gather’
COL - zbierać wodę z całego dachu ‘to gather water from the whole roof’
brać ‘take’, e.g.: brać torbę ‘to take a bag’
8.
a.
10
oblizać ‘lick’
The form without wy-: przedawać is obsolete in Polish. In all probability prze- functioned here as a prefix.
See e.g. in Doroszewski (1965) the entry for prze-, which sequence in the history of the Polish language had as
one of its meanings the change of location.
11
Again this verb has only prefixed related predicates, or the form with the reflexive clitic, e.g.: wygarniać ‘take
out’, garnąć się ‘flock around’.
COL – oblizać loda z palców (A.M.)‘to lick icecream off one’s fingers’
COS – oblizać palce z lodów (A.M.) ‘to lick fingers of icecream’
lizać ‘lick’, e.g. lizać lody ‘to lick icecream’
b. obrabować ‘rob’
COS – Obrabowali domy z kosztownosci ‘They robbed houses of precious things’
rabować ‘rob’, e.g. rabować dom ‘to rob a house’, rabować kosztowności ‘to rob precious things’
c.obierać ‘peel’
COL - obrać skórkę z limetki ‘to peel the skin off a lime’
COS - obieramy pyrkę z łupiny ‘We peel the potato from its peel’
brać ‘take’, e.g.: brać lekcje ‘to take lessons’
d. oswobodzić 12 ‘free’
COL - oswobodzić piętę z pułapki ‘to free one’s heel from a trap’
9.
a. odzyskiwać ‘rescue, recycle’
COL – odzyskująca węgiel z hałdy. ‘re-cycling coal from a pile’
zyskać ‘gain’, e.g.: zyskać zaufanie ‘to gain trust’
b. odsłaniać 13‘unveil, remove’
COL – Odkryła moją ranę, na szczęście nie odsłaniając maty z prawego ramienia. ‘She discovered my
wound, fortunately without removing the mat from my right arm’
COS - Odsłania wyspę z mgły. ‘It unveils the isle from fog’
c.odciąć ‘cut off’
COL - Nożyczkami odcinano kartkę z bloczku. ‘With a pair of scissors one cut off a page from a pad’
ciąć ‘cut’, e.g.: ciąć papier ‘to cut paper’
d. odganiać ‘chase off’
COL - odganiam wronę z balkonu ‘I am chasing a crow from the balcony’
gonić ‘chase’, e.g. gonić wiatr ‘to chase the wind’
e. odgarniać14 ‘pull away’
COL - odgarnia włosy z czoła ‘She pulls her hair away from her forehead’
f. odzyskiwać ‘rescue’
COL - odzyskiwać energię ze światła ‘to get back energy from light’
zyskać ‘gain’, e.g. zyskać uznanie ‘to gain esteem’
10.
a. oczyszczać ‘cleanse’
COS - oczyszcza duszę z jednego grzechu ‘He cleanses soul of a single sin’
czyścić ‘clean’ (see 1.a)
b. okradać ‘steal’
COS - Okradała Reginę z resztek jedzenia ‘She robbed Regina of the remains of her food’
kraść ‘steal’, e.g.: kraść pieniądze ‘to steal money’
c.ograbić ‘rob’
COS - ograbili dom z mebli ‘They robbed the house of furniture’
grabić ‘rob’ (see 1.e)
d. odzierać ‘tear’
COS - odziera świat z tajemnicy ‘to deprive the world of mistery’
drzeć ‘tear’ , e.g. drzeć papier ‘to tear paper’
e. osączyć ‘soak’
COS - osączyć rybę z zalewy ‘to soak fish of water’
f. osuszać ‘dry’
COS - osuszał czoło z potu ‘He dried his forehead from sweat’
12
Again this verb does not have non-prefixed alternants, however the nominal stem: swobod(a) ‘freedom’ is
attested.
13
There are no verbs without any prefix that would correspond to odsłaniać. However other prefixed verbs
exist in Polish with this root, e.g. przesłaniać ‘screen’, zasłaniać ‘veil’, osłaniać ‘protect’, etc.
14
No non-prefixed variants appear, see ftn. 11.
suszyć ‘make dry’, e.g. suszyć ubranie ‘to dry clothes’
11.
a. posprzątać ‘clear’
COL -posprzątać śmieci z brzegów jeziora ‘to clear trash off the water-line of a lake’
COS - posprzątali stadion z kamieni i butelek ‘They cleared the stadium of stones and bottles’
sprzątać ‘clear’(see 1.c)
b. usuwać ‘remove’
COL - usuwać bezdomnych z dworców ‘to remove the homeless from stations’
suwać ‘push’, e.g.: suwać nogami ‘to drag one’s legs’
As the data in this section show, prefixed verbal stems in Polish, even if they correspond to
non-prefixed non-alternating verbs, frequently appear in both COS and COL frames. This calls
for an explanation, which will be forthcoming in section 8.
3. Research bacground: Rappaport Hovav and Levin (1998), Levin (2006) for
English, Segal and Landau (2012) for Hebrew, Alexiadou and Anagnostopoulou
(2012) for Greek
‘Clear’ alternation poses important questions in the area of morpho-syntax. The most obvious
one is why some verbs should allow the alternation, while it is inadmissible with some others.
The next question involves scanning a more extensive language area than just one language,
and it consists in the query whether there are differences between languages with respect to
this alternation. The most basic, yet most profound one asks about the nature of lexical
representation of verbs and grammatical contribution to their morpho-syntax. Then one
should ponder over semantic and/or morphological variation among ‘clear’ verbs and its
contribution to the availability of the alternation. Leaving the last question to be taken up in
the final section of this text, here we will concentrate on the remaining issues, as a few
analyses within the area of ‘clear’ alternations are available and they will prepare the stage for
more detailed considerations relating to Polish.
Rappaport Hovav and Levin (1998) and Levin (2006) discuss the ‘clear’ alternation for
English, noticing that in English it is applicable to very few verbs: clear, clean, drain and
empty, while for some other verbs only COS or COL frames are available, e.g.:
12. Henry cleared dishes from the table. Henry cleared the table of dishes.
The thief stole the painting from the museum (COL . *The thief stole the museum of the painting.
The doctor cured Pat of pneumonia (COS). *The doctor cured pneumonia from Pat.
The linguists conclude that the availability of alternating power depends on the type of verb:
Only manner verbs can appear in the alternating structures and their lexical entries have the
following form:
13. [XACT<CLEAR>Y ]
X is an actant inciting the action, while Y is a pure root participant, which means that it
appears as an argument in the lexical representation of a verb, but it does not function in a
more extensive event structure with which this verbal representation may be supplemented by
grammar. This root participant status is manifested in the way the argument behaves in
syntactic structures; First of all it may take up various guises, e.g. of the Direct Object in the
Accusative Case or of a prepositional phrase. Alternately it may not be realized overtly, or
realized as a non-specific object.
Manner verbs have comparatively simple structure, as compared with result verbs, which
cannot alternate as they cannot be augmented due to their already cumbersome representation
as given in (14 below):
14. [[x ACT] CAUSE [BECOME [y<RES-STATE>]]]
(e.g., break, dry, melt, open, split)
Consequently, it is manner verbs, whose representation may be augmented with further
event structure, which show alternations like ‘clear’ alternation, schematically rendered in
(15):
15. [XACT<CLEAR>Y ]CAUSE[BECOME[z NOT AT<PLACE>]
e.g.:She (x) cleared dishes (z) off the table (PLACE)
[XACT<CLEAR>Y ] CAUSE [BECOME [z NOT WITH <THING/STUFF>]]
e.g.: She (x) cleared the table (z) of dishes (STUFF)
Alexiadou and Anagnostopoulou (2012) (see also Segal and Landau 2012:233) apply
Rappaport Hovav and Levin’s (1998) and Levin’s (2006) ideas to Greek, proposing that the
event frames augmenting manner verbs look as follows:
16. COL: [X CAUSE [Y BECOME[AWAY FROM Z]]]
17. COS: [X CAUSE [Z BECOME [WITHOUT Y]]]
Consequently, manner alternating verbs will be augmented by grammar through so called
Canonical realization rules (see Rappaport Hovav and Levin 1998:109) with COS and COL
frames, while the non alternating verbs, allowing just one of the alternating possibilities, will
be specified as such in the lexicon. The lexical representations of non-alternating verbs would
look (after suggestions from Levin (2006:15) and Alexiadou and Anagnostopoulou’s (2012))
along the lines in (18) and (19) below:
18. [[x ACT] CAUSE [y BECOME [ AWAY FROM z]] (with non-alternating COL verb)
19. [[x ACT] CAUSE [y BECOME [WITHOUT z]] (with non-alternating COS verb)
If the distinction critical to the (non)appearance of the alternation exists between manner and
result verbs, than the division should, we think, correlate with some other tangible distinctions
between these classes of verbs in the ‘clear’ group. For instance, these differences could be
read off verbal meanings, or result from their morphological, morpho-syntactic and syntactic
behavior, beside the ‘clear’ alternation facts. Were it not the case, the appeal of the proposal
would be significantly diminished. In the last section we will show that at least in the area of
morphology such correlatives appear in Polish. First, however, we will adumbrate some such
correlatives that Alexiadou and Anagnostopoulou (2012) (after Rappaport Hovav and Levin
1998 , Levin 2006), as well as Segal and Landau’s (2012) discuss and confront them with
Polish data.
4. Entailments with alternating and non-alternating verbs in Greek and in Polish; a
contrastive account
Alexiadou and Anagnostopoulou (2012) (after Rappaport Hovav and Levin 1998 , Levin
2006) support their analysis with distinctions in the area of entailment that can be observed
between alleged manner and result verbs. As alternating verbs do not possess in their lexical
representation all the arguments that figure in the full frame, it is believed that the truncated
structure should not entail the absent argument. This seems to work for Greek perfectly:
20. (Alexiadou and Anagnostopoulou (2012) for Greek:
Alternating verb with V DP-Stuff (No COL entailment)
Kseplina tin laspi (alla den efige apo tis skales)
Washed-I the mud-ACC (but not left-it from the stairs)
‘I washed the mud (but it stayed on the stairs)’
Alternating verb with V DP-Loc (No COS entailed)
Kseplina tis skales (alla pareminan vromikes apo laspi)
Washed-I the stairs (but remained-they dirty from mud)
‘I washed the stairs (but they remained dirty with mud)’
As the missing argument is not implicit in the truncated version, then supplementing the
sentence with the information about it should not bring about an overdose of information. The
examples above attest precisely to this state of affairs. On the other hand, the result verbs,
which show just one frame15, and thus have the particular frame specified in the lexicon,
should entail both internal arguments present in the frame. Again, the entailment facts support
this claim (Alexiadou and Anagnostopoulou 2012) as the sentences augmented with the
apparently missing arguments are unacceptable due to the surplus of contradictory
information:
21. Verbs with DP-Stuff (COL entailed))
Diegrapsa tin protasi (≠≠alla paremine sto kimeno)
‘I deleted the sentence (≠≠but it remained in the text)
22. Verbs with DP-Loc (COS entailed)
Ekkenosan to ktirio (≠≠alla pareminan kapii anthropi mesa)
‘They evacuated the building (≠≠but some people remained inside)
Precisely the same state of affairs is found for Hebrew (see Segal and Landau 2012:244).
The findings suggest the analysis wherein the alternating verbs are grammatically
augmented with an additional argument in the added event structure frame, while the nonalternating verbs include both internal arguments in their lexical representations.
The analysis is, however, not supportable for the Polish data. In Polish the entailment facts
are different. The alternating verbs and the non-alternating ones behave in the same way:
Location is always entailed, but Stuff is not. This is illustrated with the data in (23-26) below:
23. Chłopiec wyczyścił dyski zostawiwszy na nich dane (cf. 1.a, an alternating verb)‘The boy
cleaned the discs having left data on them’
24. Chłopie wyczyścił dane *zostawiwszy je na dyskach ‘The boy deleted the data having left
them of the discs’
25. Uratował dzieci *zostawiwszy je w pożarze (cf. 2.c, a COL-only verb) ‘He saved the
children having left them in the fire’
15
Many result verbs may also appear neither in COL, nor in COS frame, but this does not bear directly on our
discussion. General differences between manner and result predicates are extensively presented in Beavers
and Koontz-Garboden (2012).
26. Splądrował mieszkanie zostawiwszy w nim pieniądze ‘He robbed the flat having left there
money’ (cf.3.b, a COS-only verb)16
Thus in Polish the major division does not run between the alternating and nonalternating verbs but between those with overt Stuff and without overt Stuff, as Location
is always either overtly present or entailed. This clearly shows that the status of Stuff and
Location arguments must be different in Polish. Before we delve deeper into this problem,
let us investigate some more properties of the morpho-syntax of the relevant constructions
in Polish with the specific focus on the manner vs. result distinction
.
5. Lack of objects and non-specific objects with ‘clear’ verbs in Greek and Polish; a
contrastive analysis
Levin (2006) as well as Alexiadou and Agnostopoulou (2012) associate the possibility of
non-appearance of objects or appearance of non-specific objects with manner verbs, but
not with result verbs. This follows from the proposed lexical representation of manner
verbs, where the internal argument is represented by the pure root participant, so the
element figuring there, but highly non-specific. On the other hand, the verbs which show
the rigid behavior towards the choice of event frames, i.e. the non-alternating verbs, have
to possess in their lexical entries the full information about specific arguments realized in
specific ways. Consequently, they should not allow uses without objects or with nonspecific objects. This works for English and Greek as the above sources indicate (see also
Beavers and Koontz-Garboden 2012). To some extent it also seems to work for Polish, as
shown below for the alternating and COL-only verbs:
27. Alternating verbs:
Ta pralnia czyści dobrze ‘This laundry cleans well’ (see 1.a)
Kobiety szorują i szorują od rana do nocy ‘Women scrub and scrub form the morning till the nightfall’
(see 1.b)
Trzeba sprzątać, nie da rady. ‘ You must clean up, nothing doing’ (see 1.c.)
COL-only verbs:
Trzeba usuwać – powiedział dentysta. ‘You must have (it) pulled out – said the dentist’ (see 2.c)
Są tacy co biorą, I tacy, co nie biorą. ‘There are such people who take (bribes) and such who do not take
(bribes)’ (see 2.e)
Ratuj! ‘Help’ (see 2.b.)
The above verbs seem to behave like manner verbs as they may appear without overt objects,
while the ones in (28) below sound much worse without overt objects.
28. *Policjant otrzeźwił i odprowadził do domu.‘The policeman sobered (sb.) and walked (sb.) home’ (see 3.c)
*Opróżnili z ubrań i sprzedali. ‘They emptied (sth.) of clothes and sold it’ (see 3.d)
*Ograbili i poszli. ‘They stole (sth.) and went away’ (COS-only verb)
16
We do not intend here to go into the nature of entailment and how it precisely works with various
constructions, nor is it our purpose to analyze the role or semantics of the elements introducing the entailed
clause (e.g. and vs. but). The data show clearly that there is a difference in the behavior of the Stuff argument
and the Location one, which, we feel, could be the indicator of their different status in the linguistic system. In
the discussion on the paper held at SinFonIJa it was pointed out to me that in such examples as (26) something
else than money may have been a prey to the robbery. Neverhteless, in spite of the possibility that places may
have to be also different in (25), e.g. somebody may have saved the children from drowning in a river, an
incompatibility results, which effect is absent in (26). Consequently, the status of Location and Stuff is different
in the Polish examples, unlike in Greek or English. We understand entailment as the case where in every
conceivable situation in which it is true that p, it is true that q. The particular language construction we employ
for Polish gives clearer grammaticality judgments than the sentences with Polish equivalents of but , whose
correlates are used in the quoted sources.
As the entailment data in (23-26) show, as well as the object-less examples in (27) and (28),
there is a difference in the behavior of alternating and COL-only verbs on one side and COSonly verbs on the other.
Still another body of data can be called forth to reveal a similar split, namely a different
nature of the entailed Location with alternating and COL-only verbs and this with COS-only
verbs. The analysis in the next section will testify to the same split.
6. Path entailment in Hebrew (Segal and Landau’s 2012) and Polish; a contrastive
account
Let us first introduce briefly the situation that obtains for ‘clear’ verbs in Hebrew. They
behave in much the same way as the Greek data with respect to entailment phenomena (see
section 4 above) and availability of objects (see section 5 above). However in Hebrew, unlike
in Greek or Polish, there exist alternating verbs, COL-only ones, but no COS-only verbs.
Among alternating verbs, there exists a significant difference between COL and COS uses.
COL sentences (both alternating or not) and truncated frames do not convey grammatically
relevant affectedness (result state proposition), while COS sentences do. Segal and Landau’s
(2012) draw from this state of affairs the conclusion that this affectedness is contributed by
the event frame for COS uses, and not by the lexical representation of the verb itself, nor by
the COL frame.
Let us now illustrate how they have come to such a conclusion, using English data at the
outset for the ease of exposition. Segal and Landau (2012:242) observe that: He cleaned the
table of the crumbs results in a really clean table, while He cleaned the crumbs from the table
does not necessarily render the table completely clean. Consequently, the result state
proposition is offered for the first, and not for the second utterance.
Likewise, they observe that Location is implied in the COS frame in a different way (much
weaker) than in the COL frame. Namely, it is not present for grammatical purposes in COS;
Another phrase in a sentence cannot rely on its presence. This phenomenon is referred to as
‘path meaning’ and its presence can be observed with COL frames, but not with COS ones
(see Segal and Landau 2012:244), as the data in (29-30) show:
29. šatafti et ha-boc me-ha-midraxa el ha-kvi š
‘I washed the mud from the sidewalk to the road’
30. šatafti et ha- midraxa me- ha-boc (*el ha-kvi š)
‘Iwashed the sidewalk of the mud (*to the road)’17
Disregarding many details of Segal and Landau’s (2012) discussion, the morale of this
analysis is that in Hebrew the Location argument in the COS frame is not visible to grammar
in the same way as in COLs. We may draw from these data the conclusion that COS frames
are singled out as qualitatively different from COLs in Hebrew.
Let us check this body of data for Polish. As the material in (31) will show the situation
mirrors the Hebrew ‘path meaning’ properties:
31.
17
a.
*Obrobili mieszkanie z cennych przedmiotów i pieniędzy do worka. ‘They robbed the
apartment of precious objects and money into a sack’ (COS-only verb, see 3.e)
b.
Będą usuwać bezdomnych z dworców do noclegowni. ‘They will remove the homeless
from railway-stations to shelters’ (COL-only verb, see 2.c)
A reflection forces itself upon me; namely it may be mere linear proximity which has some role to play with
this data, but another explanation is also convincing. I will return to this issue in the final section.
c. Sprzątali śmieci z ulic do worków.’They cleaned trash from streets into sacks’ (alternating
verb, see 1.c)
d. *Sprzątali ulice ze śmieci do worków.’They cleaned streets from trash into sacks’
Location is grammatically active in COL frames, whether they appear with COL-only, or with
alternating verbs, while it is not active, shows no ‘path meaning’ in COS frames.
7. Solution and conclusion
Let us now sum up the so-far findings for Polish: Entailment in Polish shows that Location is
always entailed, while Stuff is not (see 23-26); in the case of alternating and COL-only verbs
objects can be easily dispensed with (27-28), but not so with COS-only verbs; finally,
Location is grammatically active with COL frames (‘path meaning’, see 31.b, c), but not with
COS ones (31.a, c). We would like to draw the following conclusion from the above data:
‘clear’ verbs in Polish, both alternating and COL-only, are represented in their lexical entries
with single internal arguments, unspecified for their realization (pure root participants), which
would explain the behavior of these verbs re non-specific objects or objectless constructions.
As such objectless constructions can be semantically processed, the internal argument is
understood, so it must be listed with the verb. COL-only verbs are entered in the lexicon with
the Locative frame: This explains why it is this frame that they adopt if they appear in the full
bi-argumental construction. Alternating verbs have no lexical information referring to
particular frames but grammar may augment them with the two frames, indiscriminately. The
pure root participant (in alternating and COL-only verbs) is specified for Location
information, as Location always figures in entailments of alternating and COL-only verbs.
Yet pure root participants are invisible to augmenting frames, which explains why both COS
and COL frames can augment them – otherwise there would be excessive marking of
Location in locative uses of alternating verbs, with possible adverse effect on adding a COL
frame. There might arise the question: how come Location is visible in ‘path meaning’
structures in all COL cases, but not in COSs, if it should not be visible to alternating
augmented frames altogether. I believe Location’s availability for the ‘path meaning’ in COLs
must be due to the structural adjacency of locative elements in these cases; both prepositional
locative phrases in COL structures are on the structurally adjacent levels (see: 31.c):
32.
PP
PP na chodnik
V’
V
sprzątać
z ulic
NP
śnieg
In COS structures one Location is merged as the Direct Object, so it is too deep in the
structure for the ‘path meaning’ to arise.
With COS-only verbs no pure root participants appear, thus objectless constructions are felt
to be incomplete – some lexical information is lacking. Likewise, Location here is included in
the lexical frame for merging as the Direct Object and it is not available in the ‘path
meaning’.
Consequently, we believe that the alternating ‘clear’ verbs in Polish should be represented in
the following way:
32. [XACT<czyścić>Y Location ] (e.g. for the verb czyścić ‘clean’)
The non-alternating ‘clear’ verbs will have the same basic lexical representation plus the
augmented locative frame:
33. [XACT<łamać>Y Location ] CAUSE [BECOME[Z NOT AT PLACE Y] (e.g. for the verb
łamać ‘break’)
While COS-only verbs will be deprived of the pure root participant, but will be
accompanied by the COS frame:
34. [XACT<leczyć>] CAUSE [BECOME[Z NOT WITH THING/STUFF Y]] (e.g. for the verb
leczyć ‘cure’).
Whether a verb is lexicalized in one of the three above ways does not seem predictable in Polish
on any independent grounds, e.g. thanks to the specific meaning of such verbs. The data in
(35-40) include alternating and non-alternating verbs which are virtually synonymous18:
35. Żołnierze konfiskowali (‘confiscated’) mienie ze szkoły (COL-only ) ‘Soldiers confiscated
the property from the school’ vs. Żołnierze grabili (‘rob’) szkołę z mienia/mienie ze szkoły
(alternating COS/COL) ‘Soldiers robbed the school of the property/the property from the
school’
36. Joanna zbiera (‘gather, wipe out’) wodę ze stołu. (COL-only) ‘Joanna gathers water from
the table’ vs. Joanna wyciera (‘wipe out) wodę ze stołu/stół z wody .(alternating
COL/COS) ‘Joanna wipes out water from the table/the table of water’
18
The examples in this collection are mine – A.M.
37. Joanna usunęła (‘remove’) brud z koszuli. (COL-only) ‘Joanna removed dirt from the
shirt’ vs. Joanna spłukała (‘wash out’) koszulę z brudu/brud z koszuli. (alternating
COS/COL) ‘Joanna washed out the shirt of dirt/dirt from the shirt’
38. Złodziej okradł (‘steal’) dom z pieniędzy. (COS-only) ‘The thief robbed the house of
money’ vs. Żołnierze grabili (‘rob’) szkołę z mienia/mienie ze szkoły. (alternating
COS/COL)(see 35)
39. Policjant otrzeźwił (‘sober’) mężczyznę z upojenia. (COS-only)’ The policeman sobered
the man out of intoxication’ vs. Kłusownik oswobodził (‘free’) padlinę z łyków/łyki z
padliny.19 ‘ The poacher freed the carcass from the trap /the trap of the carcass’
(alternating COL/COS) vs. Myśliwy ratował zwierzę z łyków. ‘The hunter saved the
animal from the trap’ (COL-only)
40. wyrzucać odpadki z kubła (COL-only) ‘to throw the scraps out of the bucket’ vs.
opróżniać kubeł z odpadków (COS-only) ‘to empty the bucket of the scraps’
If it really is manner and result verbs which are responsible for the erratic behavior of
‘clear’ data in Polish, we lack independent evidence to distinguish them among ‘clear’
verbs. The tests supplied by Beavers and Koontz-Garboden (2012) 20 which allow us to
differentiate these subclasses do not seem to work satisfactorily for ‘clear’ data in Polish. For
instance the selectional restriction diagnostics which they present in section 4.1 should distinguish
manner verbs from result ones: Manner verbs put tight selectional requirements on their external
arguments, but result verbs are much more flexible in this respect – they accept natural forces and
inanimates. Polish ‘clear’ data show no such distinction, as the verbs in (41) below illustrate:
41.
a. An alternating verb (manner?)
Jan (agent) czyści dywan z brudu. ‘John cleans the carpet of dirt’
Ajaks ‘Ajax’ (inanimate argument) czyści dywan z brudu.
Wiatr ‘Wind’ (natural force) czysci dywan z brudu.
b. A COS-only verb
Doktor (agent) leczy go z depresji. ‘The doctor treats him for depression’
Lekarstwo ‘The drug’ (inanimate argument) leczy go z depresji.
Wiatr ‘Wind’ (natural force) leczy go z depresji.
c. A COL-only verb
Jan (agent) łamie gałęzie. ‘John breaks branches’
Samochód ‘A car’ (inanimate argument) łamie gałęzie.
Wiatr ‘Wind’ (natural force) łamie gałęzie.
Other tests do not give indisputable results either, and the entailment and indefinite object
criteria criss-cross the alternating – non-alternating distinction, as discussed in sections 4.
and 5. above. The only regularity that seems to hold is the fact that prefixed verbs are
more frequently alternating than non-prefixed ones. We will try to account for this
regularity in the next section.
19
Prof. Cetnarowska suggests that this later alternant is possible with an argument which specifies an entity
which is less important than the trap, e.g. a carcass. The sentence would be much worse if an animal was alive
and so more ‘important’ than the trap itself. I agree completely with this suggestion. The alternations which are
the topic of this paper present still many intriguing points to tackle and this hierarchy of importance is certainly
one of them.
20
For similar tests see also e.g. Levin (2006), Rappaport Hovav and Levin (1998, 2008), Levin and Rappaport
Hovav (2010, 2005, 2006).
8. Alternating prefixed verbs in Polish
In the light of the hitherto discussion it is unclear why prefixed verbs should show any
greater propensity for entering the ‘clear’ alternation than non-prefixed ones. Recall, however,
that we have postulated that the element of Location has to be present with alternating verbs.
At the same time it is precisely the information concerning the direction (which presupposes
location as its point of reference), which constitutes the main semantic contribution of Polish
prefixes to the basic verbal meaning. For instance Szymanek (2010:172)21 describes wy- (see
6 above) as contributing the ‘meaning of direction’: ‘to bring/take something out [of
Location]’, z- (see 7. above) signifies ‘downward movement’ [from a location] (Szymanek
2010:175). O-/ob- (pp.152-153) signifies ‘a movement around a given landmark’ (see 8
above), od- (p.154) means ‘movement away from some location’ (see 9 above).
Consequently, as the element of Location is assumed to be important in the lexical description
of ‘clear’ verbs (alternating and COL-only) and the prefixes, which add the meaning of
location, automatically make their derivatives more eligible for the alternating status, or for
being lexicalized with the COL frame. This analysis is supported by the fact that alternating
and COL-only verbs predominate in the prefixed data. It is just the examples in (8. e, f, g, h, i,
j) which have additional COS-only frames, when compared with their morphologically less
complex verbal relatives. Moreover, all of these COS prefixed verbs are formed with prefix o, thus analogy may be at work in these few cases.22
We feel that the prefixation facts in Polish speak strongly in favor of the proposed
representation of ‘clear’ verbs in Polish as presented in section 7.23
21
See also Svenonius (2004) for a detailed account of spatial relations implicit in prefixes in Slavic languages.
It was pointed to me by prof. Cetnarowska that other prefixed verbs show also COS-only frames, e.g. uleczyć
‘cure’. Notice, however, that leczyć (see 3.a), the non-prefixed verb, has also the same frame, so COS is not
contributed by the prefix. In the data we have managed to collect it is only the verbs prefixed with o- which
manifest this behavior in any numbers.
22
Alexiadou and Anagnostopoulou (2012) claim that the greater productivity of the alternation with ‘clear’
verbs than in the case of other verbs with double objects is conditioned in Greek by the fact that the preposition
apo, which appears with this group of verbs, is semantically more salient, directional, than se, which also
appears with double object verbs. In Polish the preposition z, which introduces the PP in our structures, is also
non-salient semantically. It represents also non-directional meanings: e.g. ‘consisting of , made of ‘ (z drewna
’out of wood’), ‘because of’ (ze złości ‘out of anger), etc. etc. Perhaps prefixed verbs, where directionality is
strengthened by the prefix, are more salient in this respect and thus abound in the alternating possibilities.
23
References:
Alexiadou, Artemis and Elena Anagnostopoulou (2012). Manner vs. result complementarity in verbal
alternations: A view from the clear alternation. In: Proceedings of North East Linguistic
Society 42.
Beavers John and Andrew Koontz-Garboden (2012). Manner and result in the roots of verbal meaning.
Linguistic Inquiry 43: 331-369.
Doroszewski, Witold (1965). Słownik Języka Polskiego, Warszawa: PWN
Gussmann, Edmund (2007). The phonology of Polish, Oxford: Oxford University Press.
Levin, Beth (2006). English object alternations: A unified account. Unpublished Manuscript.
Levin, Beth and Malka Rappaport Hovav (2010). Lexicalized meaning and manner/result
complementarity. Ms. Stanford University and The Hebrew University of Jerusalem.
Rappaport Hovav, Malka and Beth Levin (1998). Building verb meaning. In: The Projection of
Arguments: Lexical and Compositional Factors, Miriam Butt and Wilhelm Geuder (eds.).
CSLI Publications: 97-134.
Rappaport Hovav Malka and Beth Levin (2010). Reflections on manner/result complementarity. In:
Syntax, Lexical Semantics, and Event Structure, Malka Rappaport Hovav, Edit Doron and Ivy
Sichel (eds.). Oxford: Oxford University Press: 21-38.
Przepiórkowski, Adam, Bańko, Mirosław, Górski, Rafał L. and Barbara Lewandowska-Tomaszczyk
(eds.) (2012) Narodowy Korpus Języka Polskiego. Warszawa: Wydawnictwo Naukowe PWN.
Segal, Zehavit and Idan Landau. (2012) Clear from and Clear of: The Asymmetric syntax of
detaching. Linguistic Review 29: 223-278.
Svenonius, Peter (2004). Slavic prefixes inside and outside VP. In: Nordlyd 32.2. Special issue on
Slavic prefixes, Peter Svenonius (ed.). TromsØ: CASTL: 205-253.
Szymanek, Bogdan (2010). A Panorama of Polish Word-Formation, Lublin: Wydawnictwo KUL.
Wiland, Bartosz (2011). Prefix stacking, syncretism and the syntactic hierarchy. Lingbuzz.